## SciPy 2023 BoF

This is meant to be a shared notes document. Please feel free to help us document the discussion that take place or share your thoughts directly here as well.

## Attendees

- Paige Martin
- Demitri Muna
- Jeff Wagner
-
- (total attendance ~30) -> maybe when we started, but many people came in afterwards - I'd guess 40-45. [DM]
-
-
-
-
-
-
-

## Notes

Intro: The people who have been complaining about the lack of support for OSS are now in the NASA funding office. We have solicitations out.

Prompt: You are given lots of money - What would you do to improve science?

- Get more people involved in science
- Make incentives for research software developerment in a career. Like, have tenure committees consider it.
- Don't just fund new projects, look at existing ecosystem.
- Matching funding for private-initiated open source entiries
- Can't really do this as a government entity, or at least not directly/unconditionally.
- (Though we're interested in ideas that )
- Take technical expertise of community seriously - Putting proposals in dual-anonymous reviews with folks who haven't delivered tools before is borderline insulting. (I had a concrete proposal to NASA and someone attacked it on its technical merit, even though it was actually easy)
- Review panels are seeking more technical people, and we're happy to bring you in
- Demitri - We'll be getting an unknown amount of money, but we want to think about where we could be in 5 years and pitch that vision.
- If we're dreaming big, given that publications are so important, make a platform where code is more readily pulishable and attributable. Right now we have problems because of the incentive structure. If we sidestep that by enabling people to publish in a different way, and we can watch work evolve as open source packages, and then track their contribution to scientific output.
- Answers will depend on how "unlimited" the money is - The optimal situation would be to rebuild everything now, but more realistic/finite amounts of money will need us to evolve in steps from where we are now.
- I've thought a lot about this - If you're a big OSS developer, you're unlikely to get a faculty position because you don't get grants into your department. So instead of burning the whole system down, we can provide grants for OSS development, and then universities should quickly become interested (because they want money)
- In some ways, "rebuild it all" is the easy answer. What are the harder answers?
- The system is burning itself down, we can start building the alternative now.
- PyOpenSci and JOSS are going in the direction you're describing - They're playing by some of the same rules but going in the right direction.
- In terms of supporting individual people - There are funding sources for 3 years, and that means I may have to lay off people on that timescale. That's a big problem. We can't retain talent/talented developers if we can't give stability. So grants should be longer term/consider history more.
- The gov dumps tons of money into hardware (spaceships, colliders, etc) and its maintenance. We should fund software at that level.
- You can download the software for free, you can't download the hardware.
- I hear all the time about the faculty system - This doesn't address the RSE pathway. Not all scientists want to be faculty/teach. All of these roles should be elevated to the level of faculty. If you give the people who do the work PI status, then they can argue to get permanent positions at their university. RSEs can lead a proposal. Non-PhDs can lead proposals. So instead of trying to hack faculty, let's make our own track.
- so, advocating larger pools of money to pay for longer-term salary?
- Yeah, even like 5 year terms would be an improvement. That would greatly empower non-faculty.
-
- sustained funding for professional scientific software staff and respected career paths
- funding for improving and supporting accessibility efforts
- One example: After I got my PhD, I worked at a prestigious lab with military grants. The PIs were postdocs who pulled down millions because they really executed on the grants. We had a team of 8 software engineers working on the grants. They didnt produce many papers, instead they made software. This is an example of how it can work.
- Re: Funding software as much as infrastructure - They use the term "user facility" - we should start talking about "software user facility"s. This would give developers some stability and freedom to work on various projects. So try a incubator type setup where people can move between projects.
- Demitri - I'm advocating for that actively.
- The dream is that "finding funding" wouldn't be a separate opportunity - If you write useful code, you get paid. It would remove the need to split focus and time to seek funding.
- Re: Thinking bigger - Climate science is exploding in the private sector. Most of the data is public and generated using public funds. None of those companies are at SciPy - There should be a public/private consortium that allows academics and private sector people to meet in the middle.
- It would need to be cross-agency as well. I know there are SBIRs to support developing new tools that work well.
- One thing about "research centers"/"software centers" - my situation is being funded by an NSF-funded institute. At the end of our 5 years, everything ends. so instead of just ONE project running out of support, ALL the projects in our center will run out of support.
- I'm non-USA and maintain several OSS apckages. I got out of academia and I don't want to have my funding tied to academia again.
- The federal government should consider in-kind contributions. So paid government employees could be partially tasked with OSS contributions.
- How would you ensure that these mandated contributors woudln't be a burden on maintainers?
- You'd have to assess what those employees are doing in some way. But in-kind would be way a much easier path than trying to get money out.
- A lot of people in the OSS community aren't associated with faculty or grant agencies. And they don't get acknowledgement or money - It's the people who write the papers using the tools.
- Part of research grants could be un-spendable by the recipients but would need to be passed to OSS projects.
- Keeping people engaged in the OSS development process - Keeping some PREDICATABLE level of pay going to the people making the tools. So smaller, more predicatable amounts would be great. Lots of students leave academia for this reason.
- Hand out prestige - make up awards, hand them out, it costs nothing.
- I worked outside the USA for a long time - It matters WHO the money is given to - Govt wants to give money to universtiies and coleges. In India we have (FOSS united?) - That funding goes directly to people, not institutions. The problem with normal grants is that they go to instutions instead of people - the person who won the grant for the institution could leave the next day. I know many people who have left institutions to have the freedom to work on OSS.

How easy is it to get the finding you need?

- It depends on whether the funding is coming from the private sector, a nonprofit, or the government. Private cares about return on investment. Nonprofit may be altruistic and give you a lot of freedom. The government may give money with a lot of strings attached. It'd be foolish to assume there won't be strings attached in most cases.
- It's easy to get funding to get things that your organization is specialized for - e.g. national labs - But it's hard to get funding to contribute upstream. And it's easy to get funding for new tools instead of sustaining.
- Agree
- It's incredibly hard for most people to write proposals. Some folks are really good at it but generally people aren't. The problems include not getting feedback for improvement.
- As a tip - For NSF and NASA - You can and should call the program/grant manager. They'll give you early guidance about the feasibility of the idea.
- In a few cases, it was a guessing game to figure out what solicitations were asking for, and there wasn't much money being offered. In some cases we already work with private funders - With them it's really easy and fast. But with NASA it's way more work and it's hard to figure out how to improve.
- Different agencies having different policies and that makes it even harder.
- I've written a few proposals to NASA, NSF, and CZI. CZI is _by far_ the friendliest. They have a two-stage proposal process when you write a LOI, and you don't have to do a ton of work until your one-pager is accepted.
- I've never gone for govt funding - The private sector is incredibly easy. The public is by far the easiest way to get funding - quadratic funding. People vote on what they want and the funds are distributed according to something like a public vote. And this makes you write the proposal in a way such that normal people can read it. I started the science commons initative this way.
- Sometimes we've (and the startup behind us) applied for open source projects, and we applied for SBIR 4 times (and got it on the 4th). The review panels were extremely inconsistent and it made the process seem non transparent. I know it's supposed to be anonymized but it's hard when you don't know what kind of panel you'll get.
- It takes a lot of time to put together a ~15 page proposal+budget. Universities will usually say you have 3-5% of you time to work on proposals but that's a joke... For OSS projects, you usually want to jump into it FAST, and typical review times are close to a year, which breaks momentum.
- funding for sustained ecosystem and community support and infrastructure

Grants/NASA funding opportunities

- Are these available for non-US people?
- Basically no, this has to go to institution. But you can apply via Eureka Scientific (an institution for this purpose, VERY low overhead). NumFOCUS also qualifies.
- Also, non-Americans can't get money directly, but you can get it via a project funded by a PI at an institution.
- What about a small business?
- There would be a few boxes you have to check - you have to be registered in SAM. But in that case you're probably best served by Eureka Scientific.